UniCORE is the public foundation for advanced AI.

We design the governance, write the code, and train the people who can do both. The foundation is gifted to humanity under permissive licences. The producers who maintain it earn their authority through a 30-year apprenticeship: not bought, not granted, earned.

The result is infrastructure no single company can monopolise, built and maintained by people from every background.

The future is coming. Advanced AI will shape the world ahead, whether we are ready or not. The question is who builds it, who controls it, and who benefits. UniCORE answers all three the same way: humanity does.

If you want to spend a career building advanced AI for humanity rather than for shareholders, UniCORE is the path.

A 30-year programme from apprentice to certified producer. A public foundation given away under permissive licences. No monopoly. The work belongs to everyone, including you.

---

This is UniCORE-AI, the 12-level reference architecture layer of UniCORE. The implementation pattern that satisfies the TrueAI Nine Invariants. Sister repositories: UniVERSE (the programme) and TrueAI (the immutable Foundation).

New to producing on the public gift surface? Start with UniVERSE/GETTING_STARTED.md.

---

The three pillars

UniCORE makes three structurally distinct claims. Each one closes a different failure surface. All three are required; removing any one breaks the institutional case for the whole.

Pillar 1 — Audience: Consumer AI vs Institutional AI

Today's frontier AIs are configured for consumer use: variability, creativity, conversational warmth, and personalisation are features, not bugs. That is the right design for consumer surfaces. It is the wrong design for institutional surfaces — regulated decision-making, evidence-bound work, decisions that must be defensible to a third party. UniCORE is for the institutional audience: the same input, with the same governance state, in the same vertical, produces the same decision across vendors, sessions, nodes, and years.

Pillar 2 — Truth: the TrueAI Foundation truth contract

What an AI system is permitted to claim, and what it is required to mark as unverified. The Nine Invariants of the TrueAI Foundation make explicit: TRUE means evidenced, FALSE means falsified, UNVERIFIED means the system declined to assert. No third state where the system fabricates confidence it does not have.

Pillar 3 — Inconsistency: same input → same decision

Truth without consistency is not deployable in regulated institutional settings. If the same evidenced claim produces TRUE today and UNVERIFIED tomorrow, the truth contract is hollow. The Inconsistency Problem is the third pillar. It is closed in two places by two different mechanisms:

• Machine-side: foundation consistency (UniCORE-AI 12-Level governance + governance MD-files) + vertical consistency (per-Vertical-CORE primitives).
• Human-side: 1H1C at the production layer (Singular Pairing Principle — one human, one AI Claw, one workstream, produces certified artefacts) + xH1C at the operations layer (x humans operating the deployed Solution through one substrate Claw as consistency-holding agent, with per-Level qualification in PROD).

Canonical: THE-INCONSISTENCY-PROBLEM.md. The Singular Pairing Principle doctrine lives at bryanunitek/TrueAI/docs/10001-Singular-Pairing-Principle.md.

Honest framing

No AI architecture today can guarantee 100% end-to-end consistency. Probabilistic language models sit at the application boundary; their training is controlled by their vendors, not by UniCORE; and that training changes over time. What UniCORE makes is the structural maximum consistency achievable given that external-AI dependency, with the residual inconsistency named, bounded, and auditable.

---

UniCORE AI

Multi-Level Governance Architecture for truth-anchored, human-sovereign artificial intelligence.

Author: Bryan Fred, Unitek Systems Limited, Bedford, United Kingdom First published: May 2026 Status: Public. Given, not sold. Irrevocable.

---

What Is UniCORE AI?

UniCORE AI is a truth-anchored, governance-bound, non-self-modifying, human-sovereign architecture for advanced artificial intelligence.

It is built on the TrueAI Foundation — an immutable set of principles that define what an AI system is permitted to be, and what it is not.

It is expressed through a 12-Level Governance Model, in which every level is human-authored, human-bounded, and externally defined.

---

Governance Levels

UniCORE AI uses a 12-level deterministic vertical stack (one example implementation of the conformance floor). Any AI system with 4 or more enforced governance levels uses the UniVERSE / TrueAI / UniCORE AI foundations — regardless of level names or industry.

Level	Name	Purpose
1	Truth	Factual status of claims
2	Evidence	Human-submitted supporting material
3	Verification	Cross-checking and consistency evaluation
4	Context	Jurisdictional, temporal, and situational resolution
5	Interpretation	Meaning derived strictly from evidence and context
6	Governance	Human-authored rules from MD files
7	Compliance	Application of legal, regulatory, and mission rules
8	Operations	Governed decisions about allowed actions
9	Execution	Deterministic action taking
10	Audit	Immutable, append-only logs of everything
11	Stability	Drift detection and monitoring
12	Human Governance	The sovereign level. Intentionally imperfect. Never overridden.

Truth flows upward through Levels 1–5. Governance flows downward from Levels 12–6. No level may bypass another. No level may communicate horizontally. No level may initiate its own activity.

Human Governance at Level 12 is not the top of the stack because humans are more intelligent than the system beneath them. It is the top because the system must never be above them.

---

Reference Implementation

The enterprise reference implementation uses:

• .NET 10
• DevExpress XAF (eXpressApp Framework)
• DevExpress XPO (eXpress Persistent Objects)

Prototype target: December 2026.

Machine-level reference implementations are explicitly permitted and encouraged in C++, C, Rust, Ada/SPARK, Verilog/VHDL, and other machine-level languages adopted by humans for safety-critical domains.

---

Architecture Documents

Code	Title
00005	UniCORE AI Technical Architecture Diagram (Text-Based)
00015	UniCORE AI Technical API Specification
00016	UniCORE AI Data Model (XPO + SQL Schema)
00017	UniCORE AI Inter-Level Messaging Protocol (ILMP)
00025	UniCORE AI Multi-Model Integration Framework
00027	UniCORE AI Master Document
00029	Machine-Level Integration (C++ and Other Machine Languages)
00035	UniCORE AI Technical Reference (Developer Edition)
20001	Why the Rules Don't Live in the Prompt
20002	Compaction and Durable Continuity

---

Governance Documents

Code	Title
00020	UniCORE AI Compliance & Regulatory Mapping
00021	UniCORE AI Risk Register (Enterprise + Space Mission)
00026	UniCORE AI Global Governance Charter
00032	UniCORE AI Licensing Framework
00033	UniCORE AI 100-Year Roadmap (2026–2126)
00036	UniCORE AI Governance Simulation Scenarios

---

Operations Documents

Code	Title
00018	UniCORE AI Space Mission Governance Pack
00019	UniCORE AI Enterprise Deployment Guide
00022	UniCORE AI Operational Playbook
00023	UniCORE AI Training & Certification Framework
00024	UniCORE AI Incident Library (100+ Enterprise & Space Mission Scenarios)
00044	UniCORE AI Multi-Model Robotics Governance
10001	Singular Pairing Principle (mastered in TrueAI — mirror)
10002	Certification Before Layered Governance (mastered in TrueAI — mirror)
10003	Generation IT Succession (mastered in TrueAI — mirror)

---

UniCORE Positioning Principle

The programme is positioned as Harmony, Peace, Space Exploration, for Humanity.

Industries and uses that align with this positioning are welcome. Those that do not are not. Military uses are intentionally absent from the programme and will not be added.

The "Powered by UniCORE AI" and "built on TrueAI Foundation" certifications must not appear on any military use. The badge is part of the gift, and the gift is meant for Harmony, Peace, Space Exploration, for Humanity — using the badge to brand weapons-class systems would invert the gift principle. The positioning closes that route.

This is a structural choice, not a marketing choice. The programme exists to keep critical decision systems available to humanity as gift.

---

Related Repositories

• UniVERSE — The wider vision, whitepapers, strategy, roadmap, and outreach.
• TrueAI — The immutable Foundation. Small. Stable. Never commercial.
• UniCORE — The UniCORE name and the open list of UniCORE.<Industry> classified names. CC BY 4.0 gift surface.
• UniCORE.GVB — The substrate-services layer of the programme (Global Virtual Bridge). CC BY 4.0 gift surface.
• UniSaaS.UniCORE — The SaaS-deployment-shape sister of UniCORE. Same governance, multi-tenant topology. CC BY 4.0 gift surface.
• UniSaaS.UniCORE.GVB — The SaaS-deployment-shape sister of UniCORE.GVB. Substrate-services layer for the SaaS topology. CC BY 4.0 gift surface.

The public register of practitioners certified by Unitek Systems Limited to provide Solution Review and other delegated programme services lives in UniVERSE/CERTIFIED-EXPERTS.md.

Attribution

Powered by UniCORE AI. Built on the TrueAI Foundation.

Attribution required wherever UniCORE AI, the TrueAI Foundation, or the 12-Level Governance Model is referenced, implemented, or extended.

---

Declaration

UniCORE AI is truth-anchored, governance-bound, non-self-modifying, non-self-creating, non-autonomous in authority, human-sovereign, deterministic, auditable, long-duration stable, free, public, open, non-proprietary, irrevocable, and given, not sold — released under CC BY 4.0 in LICENSE.md. Commercial use of the architecture is permitted under that licence with attribution preserved; the names and certifications are protected separately by trademark and the Brand and Trademark Use Policy.

The next century of artificial intelligence will be defined not by the power of its models, but by the integrity of its governance.

This is that governance.

— Bryan Fred, Unitek Systems Limited, Bedford, United Kingdom, May 2026.

---

Why these repositories exist

Powered by UniCORE AI built on the TrueAI foundation is a moral and ethical answer to Advanced AI for humanity.

UniVERSE, TrueAI, and UniCORE-AI are a way for humans and AI to build advanced AI applications together — under the rules, structure, and governance these three platforms set out in the open. They let humans work with AI, and communicate with other AIs, in a way that is good: one that preserves life on Earth and, in time, throughout the entire universe.

There is much humanity does not yet understand about what is coming. This is a first step.

Truth brings harmony — whether between humans, between AIs, or between the two. Harmony is peace.

This is a gift. Public. Not for sale. Irrevocable. There is no negotiation. Anything else would be false, and falsehood cannot build what this is for.

— Bryan Fred, Unitek Systems Limited, Bedford, United Kingdom

---

Discuss and contribute

Adoption questions, architecture critique, scenario stress-tests, implementation reports, and corrections belong in GitHub Discussions. See DISCUSSIONS.md for what the space is for, what it is not for, and how to take part.

Classification, brand, and claims

UniCORE AI is a governance architecture, not a product, platform, SaaS offering, tool category, or brand for sale. See STATEMENT-ON-CLAIMS.md for the binding rules on how the UniCORE AI and TrueAI names may and may not be used, including in GitHub Discussions. The canonical TrueAI-specific statement is published at TrueAI/STATEMENT-ON-CLAIMS.md.

Landscape

Where UniCORE AI and the TrueAI Foundation sit in the current AI governance landscape — a structural comparison against the published governance frameworks of OpenAI, Anthropic, Google DeepMind, Microsoft, Meta, and others. See LANDSCAPE.md for the comparison and what it shows. The canonical version is published at UniVERSE/LANDSCAPE.md.

Horizon

The UniCORE AI reference architecture is ready to cite and implement today. Full-conformance implementations and broad multi-sector adoption are a 10–20 year horizon, possibly longer. See HORIZON.md for the author's time-horizon statement. The canonical statement is published at TrueAI/HORIZON.md.

Whitepaper

The canonical architecture whitepaper is published at docs/whitepaper/WHITEPAPER.md. It depends on the TrueAI Foundation and is the artefact of record for cite-this-when-discussing-UniCORE-AI purposes. See docs/whitepaper/README.md for the index entry.

AI authorship

This repository is produced with AI assistance operating under TrueAI governance. The full disclosure is at AI-AUTHORSHIP.md.

Deployment principles

A UniCORE Solution built on the TrueAI Foundation is subject to two deployment-topology principles. They govern how Solutions must be produced and certified before layered governance (Regional, national, organisational, mission, etc.) is applied to them. They are not invariants; they sit alongside the Gift Principle. Canonical home for both is the TrueAI Foundation; mirror copies live here for in-repo reading.

• Singular Pairing Principle — one human, one AI Claw per workstream; Generation IT producer qualification; direct pairing or parallel isolation with fresh synthesis.
• Certification Before Layered Governance — a Solution must conform to the Foundation before per-level governance MD files may be applied to it. Build → Certification gate → Govern. Per-level MD authoring belongs to the Govern phase, after the gate.

Contact

• Public discussion: GitHub Discussions (see DISCUSSIONS.md)
• Private contact / connection request: LinkedIn